TextBounds

type: module path: "@root/src/PuduLangMcp/Utils/TextBounds.pudu" fidelity: Active grammar: "[[grammar/pudu]]" depth_score: 0.4 depth_status: MODERATE coupling: 0 interface_stability: 0.9 tags: [module] aliases: [Text Bounds]

Purpose

Keep text handed to a model within a size, say when it was cut, and pick the part of a passage around a match.

Interface

Signatures

export fn bounded(text: Str, limit: Int) -> Str
export fn snippet(text: Str, needle: Str, width: Int) -> Str
export fn singleLine(text: Str) -> Str

Linkage

• Requires: Std.Text.
• Consumed by: [[src/PuduLangMcp/Domain/Docs/Search]], App/Tools/*.

Algorithm

• bounded: text of at most limit characters is returned unchanged; longer text keeps its first limit characters and gains a final line [truncated: N more characters]. limit is not negative.
• snippet: find needle case-insensitively (an empty needle is no match); take width characters starting half a width before it, never before the start, prefixed and suffixed with … where text was dropped. No match answers the first width characters, with … when more follow.
• singleLine: line breaks and tabs become single spaces; runs of spaces collapse.

Negative Logic (Prohibited Paths)

• bounded never answers more than limit characters plus its marker line.

Edge Cases

• A limit of 0 answers only the marker, and empty text within it answers empty text.

Depth

DEPTH 0.4.

Grill Log

• Q: Characters or bytes? A: Characters; the cap protects a model's context, which is counted in text, and a byte cut could split a character. Byte caps belong to process capture.

Referenced by

[[src/PuduLangMcp/Utils/_MOC]]